I agree Xanadu is a nice idea. But it's also a classic chimera -- an impossible world.
As I understand it, Xanadu is built around transclusions, or two-way semantic links, (i.e. verbs) between paragraphs, photos, graphs, etc (i.e. nouns). Of course verbs do not flow equally well in both directions between subject and object, nor along a hierarchy between super-objects and derived-objects. Nor do nouns remain static. As they are edited, their category will change, invalidating some of the attendant links.
How to correct this? Inject context (attach semantic categories to each link and each node). Yeah. Right.
Xanadu would be severely unmanageable. The explosion of small links and contextual tags would rapidly overwhelm the system. Then due to inaccurate or insufficient landmarks, link navigation would become downright Rorschachian -- a journey punctuated by many false starts down blind alleys, often failing to reach the intended destination. I suspect it'd be a lot like the thought process of someone with ADD. Like Ted Nelson.
So in this perfect world, we should tax your actions if they incur costs to others. Hmm, therefore we should tax:
- people who would pass any new law (because more law enforcement costs more money)
- people who oppose eliminating any old law (because... see above)
So before government is entirely defunded by the enforcement of this law, in order to pay for it we should quickly tax those who voted for this law. After that, we'll tax everyone who ever voted to pass any other law.
Whee! The government is rich again! And it's out of a job too!
I've been trying to understand the appeal of those very oxymoronic tenets that most Americans have embraced since Reagan's reign. The best source of perspective I've found so far has been "What's the Matter With Kansas?" http://en.wikipedia.org/wiki/What's_the_Matter_with_Kansas%3F. It's an engaging look at the rise and wholesale adoption of those very mindless platitudes.
Antisyzygy didn't propose that we kill all the lawyers (yet), but that America values interlopers and middlemen like lawyers and MBAs over those who produce the intellectual property that advances and sustains modern business. To that I say, ABSOLUTELY. No lawyer or MBA every invented any widget of any value or built a great idea into a great company. At best, these two professions grease the wheels of commerce. And these days, US business needs more grit and less grease.
The very *last* thing America needs is more middlemen. We need more fresh ideas, clever inventions, and daring folks with the courage to break glass ceilings. Inventor, come. Bureaucrat, go.
Given that foreign competitors already take away a large fraction of new tech opportunities via off-shoring, you propose we should encourage them to come here, get H1Bs, and then *also* take away on-shore tech opportunities? We should design the H1B program to better eliminate both off *and* on-shore tech jobs?
I think Dr Rumelhart should be remembered as a true founder of AI. While he wasn't there at the beginning (Dartmouth 1954), his work with McLelland, Hinton and Williams resurrected not just neural nets but in many ways the entire field of AI.
In 1969, Marvin Minsky and Seymour Papert published "Perceptrons" which emphasized the inadequacy of simple single layer NNs and effectively discouraged further funding of NNs, thereby directing the mainstream of AI research monies into symbols and logic for the next 20 years. But by the mid 1980s, it had become clear that AI was not living up to the promises of its leading lights to quickly produce a thinking machine. It was Dr Rumelhart (among others like Grossberg) who further investigated and developed NNs, integrating novel reinforcement techniques (e.g. backpropagation) and thus grounding the field of AI more mathematically, or as it was also known then, subsymbolically.
Since Dr Rumehart's work in the 1980s, subsymbolic AI has risen in importance as symbolic AI has fallen. Today, virtually all of AI research employs engineering techniques governed by increasingly sophisticated mathematical principles, and integrate feedback and learning, just as his work did. While I wouldn't claim Dr Rumelhart to be the father of modern AI, I would point out that the mathematics and machine learning central to his work correctly anticipated the current grounding of AI in both learning and numerical computation that reshaped and resurrected the field just as symbolic AI degenerated into the the "AI Winter" of the 1980s. Today's numerical AI researcher resembles subsymbolicists like Rumelhart significantly more than the renowned founders of AI, symbolicists all.
From what I'm hearing, Dr Sejnowski's plaint only partly addresses the problem. To implement cognition using a computational model, we need a neural simulator that:
- is large enough to represent all the neurons and interconnections needed to synthesize human-level cognition - uses much less power than a supercomputer
But to be more than "a brain in a jar" it also must:
- learn using supervised and unsupervised instruction - quickly load and unload modules of what it has learned
Without addressing all four goals, you've just recreated what we already have with its inherent limitations:
- an abstraction that is clunky and inefficient -- neural nets (analog) run on von Neuman (digital) architectures - an implementation that uses too much power (supercomputer) - knowledge representation that is not modular or decomposable
So even if we can devise a more appropriate implementation than NNs on the Tianhe-1A, we're still a long way from nirvana.
1) it can parse language well, even with embedded puns, and identify the fact sought in the question
2) it can navigate a large amount of general domain info and match the context (i.e. frame) of the question to the missing fact.
This is admirable, BUT NOT GENERALLY USEFUL. Watson may make a fine reference librarian, but it is not 'learning' (except for building association weights among potential answers, probably using a fairly simple measure of 'relevance'). It is not deducing, inferring, learning, planning, or resolving, and that's going to make for a damned poor auto mechanic, much less uber-doktor.
Remember when Deep Blue beat/tied Kasparov? Did you notice what revolutionary IBM products followed from that? Not a one.
Watson was a fun demo that made IBM look good. That's all.
I agree on both counts. Not knowing its history, I saw Watchmen cold and was amazed and fascinated by its evocative characterizations and its insightful twists on the age-old DC comic superhero zeitgeist.
If the measure of a movie is how much it affects you, then Watchmen succeeded with me as no film has for 20 years.
Replicating Watson's software will not necessarily be trivial. If the target machine's CPU is not IBM's POWER7 cell processor, I definitely would not just assume that the method of indexing the data will translate as efficiently, or that the memory bandwidth will be close to comparable, or that the algorithms will scale linearly onto a different number of cores.
In general, optimized parallel codes are very hard to rescale onto a different number of processes, much less a new architecture. The POWER7 has a very high data transfer rate, unlike commodity PC CPUs. If you dropped Watson's software onto a conventional cluster of the same size, I suspect it would run 10 times more slowly. If you also cut the machine from 2880 cores and 15 TB to a more conventional 64 cores and 50 GB, Watson's runtime would go up more than linearly. And if Watson's code did scale linearly onto such a machine, its.5 second response time should rise (2880/64)/10 fold, to about 220 seconds per question.
Hmm. Maybe they should play the Jeopardy theme while you wait for your answer...
I disagree. Most phone support questions are not resolved with a one word answer (or even a short sentence). They usually include specific instructions on how to explore and diagnose a problem and then how to fix it, both of which are multi-step drill-downs through a hiearachy of menus. So far, Watson has showed none of these skills.
Nor did Watson show that it can do diagnosis, which is the application area for it that IBM is targeting first (e.g. physician's assistant). This requires not only deduction but a sense of time (this came before that), and induction (give and take Q&A), and resolution, which Watson also didn't demonstrate. Watson is a whiz at bayesian reasoning, but that only takes you so far in the real world.
Watson showed that computers can parse natural language well; it can understand the gist of a question (the "open slot") and resolve that uncertainty. And it's astonishingly fast. I give IBM a ton of credit for the implementation, especially for encompassing so much information and beating the best humans at a tough task.
But it's not at all clear that Watson's current set of abilities will scale to other uses like medical diagnosis. After all, 2880 processors and 15 terabytes of memory makes for a very expensive solution to ANY problem.
As kellyb9 suggests, a front panel free of buttons is likely to appeal to Jobs & Co. But instead of multitouch, it'd be easy and preferable to reset using a second button on the side, like volume.
Resetting via multifinger multitouch makes no sense, nor would simplifications like one finger pressing repeatedly, or one finger pressing and holding.
Occasionally iPhones freeze and ignore further screen input. What then?
A physical-button-based reset is essential on any device vulnerable to the Halting Problem.
Fact is, software development has relied on exponential hardware speedup for the last 40 years, and that's why Moore's Law *is* still relevant.
If a global computer speed limit is nigh then mainstream computing will slowly decelerate. Why? 1) Perpetual increase of bloat in apps, OSes, and programming languages. 2) Ever more layers of security (e.g. data encryption and the verification & validation of function calls, data structures, and even instructions). 3) Increasing demands of interactivity (e.g. event polling of network & GUI & the explosion of sensors).
The care, feeding, and survival and all of this crap depends on a nonstop increase in hardware horsepower. If hardware no longer improves (regardless of whether it was due to transistor density, clock speed, or microarchitecture design), the effect is the same -- computers won't keep up with the rising demands on them, and they will slow down.
In fact, Moore's Law died the minute that multicore CPUs were announced as its heir apparent. Parallelism has always been the last ditch solution to hardware speedup. (Amdahl's Law is a harsh mistress). Parallel CPUs have been available forever and were mercifully avoidable as long as clock speeds multiplied. But when CPU Hz topped out at around 3 billion (around 2003), the hardware pros knew the party was over.
With more people than any other country, obviously china will deliver more of their language than will any other nation. But will any other nation speak chinese? No.
As the economic iinfluence of the japanese rose during the 1980s, pundits foretold of the rise of all things japanese. They were wrong, not just in outcome but in premise.
It isnt economic power alone that influences and pervades world culture, its innovation and ingenuity. The US had these things for the past 150 years, largely due to the economic and political opportunities this culture afforded to those insanely motivated immigrants who could not break through the glass ceilings in their home country. That culture most certainly does not exist in china, not now, not ever before, and not for the forseeable future.
TFA confuses supply with demand. Today's China is all about supply, but is a great big zero when it comes to demand.
Make that $1000 for a genome NEXT year, not 2015. The current cost for a genome (at maybe 5x ocersample) from the Beijing Genomics Institute is about $5000 in bulk, and dropping fast. Speed, reagent cost, oversampling, and completeness (e.g. over 95%) are all improving at super linear rates.
For more, read "The $1000 Genome", printed in mid-2010. It makes clear that gene sequencing technology is The Next Big Thing, and imminent. The question that remains is, how useful will the info prove to be until tech like gene therapy is workable. For now, genomic data is almost entirely a novelty -- mostly good for entertainment value.
If Google's change does what's intended, downrank URLs of merchants who invite furious web opinion as a marketing ploy to game search engines, only the losers, like bile-thriving DecorMyEyes' Vitaly Borker will seek alternative means of self promotion. Frankly, I suspect that's a pretty small contingent of potential astroturfers -- 'hundreds' according to the Google blog.
I suspect that Google has indeed applied Sentiment Analysis, but done so narrowly, targeting only (1)merchants described with (2)domain-specific epithets and phrases from (3)buyers who are unhappy.
BTW, here's the link to today's NYT article. It and Sunday's original article are an worthwhile read.
Right. I'm pretty sure the underlying technology is based on Seam Carving , where a continuous background region is collapsed 'seamlessly' (or an object within such a region is removed). This doesn't work so well when the background is discontinuous, so it's not going to remove logos from clothing. It's also not going to work well on live video, since the object to be removed needs to be identified manually before the excision can occur. But it works nicely on prerecorded media. I'm impressed at how well it works on a brick background.
It's _religion_ that causes confusion between reality and fiction, not technology. The pope mistakenly equates what you see with what you believe. Just because you see something that looks real doesn't mean you believe it's real. And just because you believe something is real doesn't make it real.
Not surprising that a religiously orthodox type like the pope would have trouble with subtleties like these, really.
Use two monitors: one horizontal, one vertical. Then put the right app on the right window. 'Nuff said.
And BTW, the reason that many LCD monitors look like crap when rotated is... it's a cheap monitor. Twisted Nematic (TN) TFT LCD displays are the standard in low-to-mid-range laptops. They undergo drastic acromaticity when the viewing angle shifts more than a few degrees.
So if you want to rotate your monitor, buy one that uses a better LCD technology (like IPS, AFFS, MVA, PVA, etc).
First I would lay out your data using meaningful labels, like a directory named for the acquisition date + machine + username. Never change this. It will always remain valid and allow you to later recover the data if other indexes are lost. Then back up this data.
Next build indexes atop the data that semantically couple the components in the ways that are meaningful or acessible. This may manifest as indexed tables in a relational database, duplicate flat files linked by a compound naming convention, unix directory soft links, etc.
If you're processing a lot of data, your choice of indexes may have to optimize your data access pattern rather than the data's underlying semantics. Optimize your data organization for whatever is your weakest link: analysis runtime, memory footprint, index complexity, frequent data additions or revisions, etc.
In a second repository, maintain a precise record of your indexing scheme, and ideally, the code that automatically re-generates it. This way you (or someone else) can rebuild lost databases/indexes without repeating all your design and data cleansing decisions, and domain expertise. This info is often stored in a lab notebook (or nowadays in an 'electronic lab notebook').
I'd emphasize that if you can't remember how your data is laid out or pre-conditioned, your analysis of it may be invalid or unrepeatable. Be systematic, simple, obvious, and keep records.
Slashdot Mirror
I agree Xanadu is a nice idea. But it's also a classic chimera -- an impossible world.
As I understand it, Xanadu is built around transclusions, or two-way semantic links, (i.e. verbs) between paragraphs, photos, graphs, etc (i.e. nouns). Of course verbs do not flow equally well in both directions between subject and object, nor along a hierarchy between super-objects and derived-objects. Nor do nouns remain static. As they are edited, their category will change, invalidating some of the attendant links.
How to correct this? Inject context (attach semantic categories to each link and each node). Yeah. Right.
Xanadu would be severely unmanageable. The explosion of small links and contextual tags would rapidly overwhelm the system. Then due to inaccurate or insufficient landmarks, link navigation would become downright Rorschachian -- a journey punctuated by many false starts down blind alleys, often failing to reach the intended destination. I suspect it'd be a lot like the thought process of someone with ADD. Like Ted Nelson.
So in this perfect world, we should tax your actions if they incur costs to others. Hmm, therefore we should tax:
- people who would pass any new law (because more law enforcement costs more money)
- people who oppose eliminating any old law (because... see above)
So before government is entirely defunded by the enforcement of this law, in order to pay for it we should quickly tax those who voted for this law. After that, we'll tax everyone who ever voted to pass any other law.
Whee! The government is rich again! And it's out of a job too!
And my $10 hammer can beat your $200 laptop.
Amen brother!
I've been trying to understand the appeal of those very oxymoronic tenets that most Americans have embraced since Reagan's reign. The best source of perspective I've found so far has been "What's the Matter With Kansas?" http://en.wikipedia.org/wiki/What's_the_Matter_with_Kansas%3F. It's an engaging look at the rise and wholesale adoption of those very mindless platitudes.
Antisyzygy didn't propose that we kill all the lawyers (yet), but that America values interlopers and middlemen like lawyers and MBAs over those who produce the intellectual property that advances and sustains modern business. To that I say, ABSOLUTELY. No lawyer or MBA every invented any widget of any value or built a great idea into a great company. At best, these two professions grease the wheels of commerce. And these days, US business needs more grit and less grease.
The very *last* thing America needs is more middlemen. We need more fresh ideas, clever inventions, and daring folks with the courage to break glass ceilings. Inventor, come. Bureaucrat, go.
Given that foreign competitors already take away a large fraction of new tech opportunities via off-shoring, you propose we should encourage them to come here, get H1Bs, and then *also* take away on-shore tech opportunities? We should design the H1B program to better eliminate both off *and* on-shore tech jobs?
What sense does that make?
I think Dr Rumelhart should be remembered as a true founder of AI. While he wasn't there at the beginning (Dartmouth 1954), his work with McLelland, Hinton and Williams resurrected not just neural nets but in many ways the entire field of AI.
In 1969, Marvin Minsky and Seymour Papert published "Perceptrons" which emphasized the inadequacy of simple single layer NNs and effectively discouraged further funding of NNs, thereby directing the mainstream of AI research monies into symbols and logic for the next 20 years. But by the mid 1980s, it had become clear that AI was not living up to the promises of its leading lights to quickly produce a thinking machine. It was Dr Rumelhart (among others like Grossberg) who further investigated and developed NNs, integrating novel reinforcement techniques (e.g. backpropagation) and thus grounding the field of AI more mathematically, or as it was also known then, subsymbolically.
Since Dr Rumehart's work in the 1980s, subsymbolic AI has risen in importance as symbolic AI has fallen. Today, virtually all of AI research employs engineering techniques governed by increasingly sophisticated mathematical principles, and integrate feedback and learning, just as his work did. While I wouldn't claim Dr Rumelhart to be the father of modern AI, I would point out that the mathematics and machine learning central to his work correctly anticipated the current grounding of AI in both learning and numerical computation that reshaped and resurrected the field just as symbolic AI degenerated into the the "AI Winter" of the 1980s. Today's numerical AI researcher resembles subsymbolicists like Rumelhart significantly more than the renowned founders of AI, symbolicists all.
From what I'm hearing, Dr Sejnowski's plaint only partly addresses the problem. To implement cognition using a computational model, we need a neural simulator that:
- is large enough to represent all the neurons and interconnections needed to synthesize human-level cognition
- uses much less power than a supercomputer
But to be more than "a brain in a jar" it also must:
- learn using supervised and unsupervised instruction
- quickly load and unload modules of what it has learned
Without addressing all four goals, you've just recreated what we already have with its inherent limitations:
- an abstraction that is clunky and inefficient -- neural nets (analog) run on von Neuman (digital) architectures
- an implementation that uses too much power (supercomputer)
- knowledge representation that is not modular or decomposable
So even if we can devise a more appropriate implementation than NNs on the Tianhe-1A, we're still a long way from nirvana.
Watson demoed two skills:
1) it can parse language well, even with embedded puns, and identify the fact sought in the question
2) it can navigate a large amount of general domain info and match the context (i.e. frame) of the question to the missing fact.
This is admirable, BUT NOT GENERALLY USEFUL. Watson may make a fine reference librarian, but it is not 'learning' (except for building association weights among potential answers, probably using a fairly simple measure of 'relevance'). It is not deducing, inferring, learning, planning, or resolving, and that's going to make for a damned poor auto mechanic, much less uber-doktor.
Remember when Deep Blue beat/tied Kasparov? Did you notice what revolutionary IBM products followed from that? Not a one.
Watson was a fun demo that made IBM look good. That's all.
I agree on both counts. Not knowing its history, I saw Watchmen cold and was amazed and fascinated by its evocative characterizations and its insightful twists on the age-old DC comic superhero zeitgeist.
If the measure of a movie is how much it affects you, then Watchmen succeeded with me as no film has for 20 years.
Or... "Press ANY key for Yes" :-)
I agree. Accurate speech recognition, especially in the presence of background noise, would be immensely more useful than Watson.
But less cool...
Replicating Watson's software will not necessarily be trivial. If the target machine's CPU is not IBM's POWER7 cell processor, I definitely would not just assume that the method of indexing the data will translate as efficiently, or that the memory bandwidth will be close to comparable, or that the algorithms will scale linearly onto a different number of cores.
In general, optimized parallel codes are very hard to rescale onto a different number of processes, much less a new architecture. The POWER7 has a very high data transfer rate, unlike commodity PC CPUs. If you dropped Watson's software onto a conventional cluster of the same size, I suspect it would run 10 times more slowly. If you also cut the machine from 2880 cores and 15 TB to a more conventional 64 cores and 50 GB, Watson's runtime would go up more than linearly. And if Watson's code did scale linearly onto such a machine, its .5 second response time should rise (2880/64)/10 fold, to about 220 seconds per question.
Hmm. Maybe they should play the Jeopardy theme while you wait for your answer...
I disagree. Most phone support questions are not resolved with a one word answer (or even a short sentence). They usually include specific instructions on how to explore and diagnose a problem and then how to fix it, both of which are multi-step drill-downs through a hiearachy of menus. So far, Watson has showed none of these skills.
Nor did Watson show that it can do diagnosis, which is the application area for it that IBM is targeting first (e.g. physician's assistant). This requires not only deduction but a sense of time (this came before that), and induction (give and take Q&A), and resolution, which Watson also didn't demonstrate. Watson is a whiz at bayesian reasoning, but that only takes you so far in the real world.
Watson showed that computers can parse natural language well; it can understand the gist of a question (the "open slot") and resolve that uncertainty. And it's astonishingly fast. I give IBM a ton of credit for the implementation, especially for encompassing so much information and beating the best humans at a tough task.
But it's not at all clear that Watson's current set of abilities will scale to other uses like medical diagnosis. After all, 2880 processors and 15 terabytes of memory makes for a very expensive solution to ANY problem.
Riiiight, nothing new or interesting in the iPad. No innovation in the Chevy Volt or the Nissan Leaf either.
I'll grant you, Microsoft can make and market a better toy. But they can't invent one. "Microsoft Innovation" is an oxymoron.
If you recall, the Kinect was invented at Carnegie Mellon three years ago by a grad student who _later_ went to work at Microsoft:
http://en.wikipedia.org/wiki/Johnny_Lee_(computer_scientist)
Oops, Johnny Chung Lee left Microsoft already. He just went to Google:
http://procrastineering.blogspot.com/2011/01/hi-google-my-name-is-johnny.html
Yet another high profile exit from Microsoft, I guess.
As kellyb9 suggests, a front panel free of buttons is likely to appeal to Jobs & Co. But instead of multitouch, it'd be easy and preferable to reset using a second button on the side, like volume.
Resetting via multifinger multitouch makes no sense, nor would simplifications like one finger pressing repeatedly, or one finger pressing and holding.
Occasionally iPhones freeze and ignore further screen input. What then?
A physical-button-based reset is essential on any device vulnerable to the Halting Problem.
Fact is, software development has relied on exponential hardware speedup for the last 40 years, and that's why Moore's Law *is* still relevant.
If a global computer speed limit is nigh then mainstream computing will slowly decelerate. Why? 1) Perpetual increase of bloat in apps, OSes, and programming languages. 2) Ever more layers of security (e.g. data encryption and the verification & validation of function calls, data structures, and even instructions). 3) Increasing demands of interactivity (e.g. event polling of network & GUI & the explosion of sensors).
The care, feeding, and survival and all of this crap depends on a nonstop increase in hardware horsepower. If hardware no longer improves (regardless of whether it was due to transistor density, clock speed, or microarchitecture design), the effect is the same -- computers won't keep up with the rising demands on them, and they will slow down.
In fact, Moore's Law died the minute that multicore CPUs were announced as its heir apparent. Parallelism has always been the last ditch solution to hardware speedup. (Amdahl's Law is a harsh mistress). Parallel CPUs have been available forever and were mercifully avoidable as long as clock speeds multiplied. But when CPU Hz topped out at around 3 billion (around 2003), the hardware pros knew the party was over.
Moore's Law, RIP.
With more people than any other country, obviously china will deliver more of their language than will any other nation. But will any other nation speak chinese? No.
As the economic iinfluence of the japanese rose during the 1980s, pundits foretold of the rise of all things japanese. They were wrong, not just in outcome but in premise.
It isnt economic power alone that influences and pervades world culture, its innovation and ingenuity. The US had these things for the past 150 years, largely due to the economic and political opportunities this culture afforded to those insanely motivated immigrants who could not break through the glass ceilings in their home country. That culture most certainly does not exist in china, not now, not ever before, and not for the forseeable future.
TFA confuses supply with demand. Today's China is all about supply, but is a great big zero when it comes to demand.
A WSJ blogger? Are you kidding me? Has Slashdot fallen so far that now you're promoting for-profit hacks like this guy?
For shame. This article is 100% unrecyclable trash.
Make that $1000 for a genome NEXT year, not 2015. The current cost for a genome (at maybe 5x ocersample) from the Beijing Genomics Institute is about $5000 in bulk, and dropping fast. Speed, reagent cost, oversampling, and completeness (e.g. over 95%) are all improving at super linear rates.
For more, read "The $1000 Genome", printed in mid-2010. It makes clear that gene sequencing technology is The Next Big Thing, and imminent. The question that remains is, how useful will the info prove to be until tech like gene therapy is workable. For now, genomic data is almost entirely a novelty -- mostly good for entertainment value.
If Google's change does what's intended, downrank URLs of merchants who invite furious web opinion as a marketing ploy to game search engines, only the losers, like bile-thriving DecorMyEyes' Vitaly Borker will seek alternative means of self promotion. Frankly, I suspect that's a pretty small contingent of potential astroturfers -- 'hundreds' according to the Google blog.
I suspect that Google has indeed applied Sentiment Analysis, but done so narrowly, targeting only (1)merchants described with (2)domain-specific epithets and phrases from (3)buyers who are unhappy.
BTW, here's the link to today's NYT article. It and Sunday's original article are an worthwhile read.
That's just silly. How can anyone (political or not) speak to a large audience without paying a mass media business to distribute the message?
Who's going to host the wikileaks website other than a business? A university? A rich patron? The Church?
Or should they rely on the largess of free opinion sites, like /.?
Right. I'm pretty sure the underlying technology is based on Seam Carving , where a continuous background region is collapsed 'seamlessly' (or an object within such a region is removed). This doesn't work so well when the background is discontinuous, so it's not going to remove logos from clothing. It's also not going to work well on live video, since the object to be removed needs to be identified manually before the excision can occur. But it works nicely on prerecorded media. I'm impressed at how well it works on a brick background.
It's _religion_ that causes confusion between reality and fiction, not technology. The pope mistakenly equates what you see with what you believe. Just because you see something that looks real doesn't mean you believe it's real. And just because you believe something is real doesn't make it real.
Not surprising that a religiously orthodox type like the pope would have trouble with subtleties like these, really.
Use two monitors: one horizontal, one vertical. Then put the right app on the right window. 'Nuff said.
And BTW, the reason that many LCD monitors look like crap when rotated is... it's a cheap monitor. Twisted Nematic (TN) TFT LCD displays are the standard in low-to-mid-range laptops. They undergo drastic acromaticity when the viewing angle shifts more than a few degrees.
So if you want to rotate your monitor, buy one that uses a better LCD technology (like IPS, AFFS, MVA, PVA, etc).
First I would lay out your data using meaningful labels, like a directory named for the acquisition date + machine + username. Never change this. It will always remain valid and allow you to later recover the data if other indexes are lost. Then back up this data.
Next build indexes atop the data that semantically couple the components in the ways that are meaningful or acessible. This may manifest as indexed tables in a relational database, duplicate flat files linked by a compound naming convention, unix directory soft links, etc.
If you're processing a lot of data, your choice of indexes may have to optimize your data access pattern rather than the data's underlying semantics. Optimize your data organization for whatever is your weakest link: analysis runtime, memory footprint, index complexity, frequent data additions or revisions, etc.
In a second repository, maintain a precise record of your indexing scheme, and ideally, the code that automatically re-generates it. This way you (or someone else) can rebuild lost databases/indexes without repeating all your design and data cleansing decisions, and domain expertise. This info is often stored in a lab notebook (or nowadays in an 'electronic lab notebook').
I'd emphasize that if you can't remember how your data is laid out or pre-conditioned, your analysis of it may be invalid or unrepeatable. Be systematic, simple, obvious, and keep records.